I'm an enterprise GCP customer, Google Support have a few superbly irritating habits:
1. They link to generic documentation that doesn't solve my problem
2. They insist that things that are clearly bugs aren't bugs until they're provided with some trivial reproduction case that satisfies them
3. They refuse to advise on issues with beta products despite half of GCP's products being in a beta
4. They are sometimes just flat-out wrong (but confidently so) about the cause of an issue
I'm an Enterprise AWS customer(well, I work for one), our account is special enough that we've got extra AWS tech people on site regularly in addition to normal support, and I've seen 1, 2, and 4 with AWS, both regular support and our on-top handholding.
On top of that AWS documentation is often both needlessly opaque, elliptical, incomplete, and outdated or otherwise incorrect.
On the other hand, if the issue isn't too obscure, AWS’s huge marketshare means that you can usually find decent answers on SO.
I am currently in a support role at a company(not GCP) with some pretty big enterprise customers and let me just say that supporting software is fucking hard.
You essentially have to be a systems engineer/sysadmin for every one of your customers with only as much context as your customer is willing to share.
You are 100% right, but it's worth noting that this is why there's a whole model of b2b interaction that involves embedded engineers partnering with a company to solve their specific problems (IBM, SAP, etc.).
The reason Google support has a hard time of it is Google doesn't offer that model to all its customers, just the ones that can pay a lot of money. But support still has the job of helping everyone else (with all the challenges you've described that such entails).
I think most of the annoyance with Google is their culturally not seeming to understand that model even exists.
Their zeitgeist appears to be "Other companies needed to provide that because their technology was wrong / incomplete. We'll just build things right instead."
Which is batshit insane, in the same way that expecting a veterinary pharmacist could prescribe for humans... with better technology.
To be fair I too probably wouldn't accept something as a bug until provided by a trivial reproduction case. I have seen too many developers call things bugs before looking at their own code.
It was a documentation/error response bug in their API, so it was hard to reproduce trivially. I gave them some source that reproduced it (https://github.com/samcgardner/neg-lb-initialiser), which really seems like it ought to be sufficient - it reliably reproduces wrong behaviour.
Give me AWS support any day